METHODS AND APPARATUS FOR IMPROVING UNIFIED SPEECH AND SOUND DECODING Russian patent published in 2024 - IPC G10L19/02 

FIELD: physics.

SUBSTANCE: invention relates to computer engineering for processing audio data. Technical result is achieved by steps of: decoding the LSF sets for each subframe from the bit stream; and converting decoded sets of LSF into representations of linear spectral pairs (LSP) for further processing, wherein subsequent processing includes determining LPC coefficients based on LSP representations by applying a root search algorithm, and wherein the application of the root search algorithm includes scaling the coefficients of the LSP representations in the root search algorithm in order to avoid overflow in the fixed point range, wherein the application of the root search algorithm includes searching for the polynomial F1(z) and/or F2(z) from the LSP representations by expanding the corresponding product polynomials, wherein scaling is performed as scaling of polynomial coefficients by degree 2, and wherein scaling includes a bit shift operation to the left.

EFFECT: reduced computational complexity when decoding an encoded MPEG-D USAC bitstream.

13 cl, 10 dwg